Monday, 20 June 2016
Alta-Deer Valley (Sheraton Salt Lake City Hotel)
Low-frequency fluctuations, determined by the unsteadiness of the mean flow, typically affect the estimate of turbulence statistical moments, especially when data are collected in complex terrain sites. In fact these fluctuations, even if characterised by small time scales, cannot be regarded as surface-layer turbulence. Therefore, removal of the average or of the linear trend from the investigated period can be not enough to separate the turbulence signal from the mean motion components. In this framework, digital recursive filters proved to be a powerful tool to identify the low-frequency components affecting the measured signals. In this contribution an improved version of the first-order digital recursive filter proposed by McMillen (1988) is disclosed. The effectiveness of this procedure for the detection of non-turbulent fluctuations was tested on a dataset of wind speed and air temperature collected on the valley floor of the Adige valley (Central Italian Alps). Data were measured at the sampling frequency of 50 Hz by means of a Gill HS Research sonic anemometer, installed at 8 m above the ground, during short field campaigns from 2000 to 2002. After removal of the block average and of the linear trend for 30-min time windows, the filter constant was evaluated on the basis of the analysis of the autocorrelation functions. Then the filtering procedure was applied and the second-order moments were estimated from the obtained turbulence signals. In particular, the stationarity of the data over the investigated periods was tested and a local similarity approach was applied in order to evaluate the thickening of the dimensionless second-order moments on the similarity relationships. Finally, as a further comparison, the previous analyses were conducted on the same dataset but without applying the filtering procedure. As a result, the reliability of the proposed filtering procedure was assessed by the improvement in the stationarity of the turbulence signals and by the considerable reduction in the scatter of the second-order dimensionless moments around the similarity functions.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner